Organization Setup
Roll out pair across multiple teams with custom KB packaging, distribution, and organization-wide standards.
Deploy pair across your organization. By the end of this guide, you will have a custom Knowledge Base package that can be distributed to every team, ensuring consistent AI-assisted development practices organization-wide.
Before starting, complete the general quickstart — install pair-cli and the Knowledge Base.
Set Up the Central Knowledge Base
Start with the default pair Knowledge Base in your central repository, then customize it:
Organization-Specific Guidelines
Add or modify files in .pair/knowledge/guidelines/ to reflect your organization's standards:
- Code design — naming conventions, design patterns, review criteria
- Testing strategy — coverage targets, testing frameworks, test categories
- Security — authentication patterns, data handling, compliance requirements
- Architecture — approved patterns, integration standards, API conventions
Standard Adoption Templates
Pre-fill .pair/adoption/ with organization-wide defaults:
- Tech stack — approved languages, frameworks, and libraries
- Infrastructure — approved cloud providers, CI/CD pipelines
- Way of working — branching strategy, review process, quality gates
Teams can extend these defaults with project-specific decisions. The adoption registry uses add behavior — it never overwrites existing decisions, so team customizations are preserved.
Package Your Custom KB
Once your KB is ready, package it for distribution:
This creates a distributable ZIP with a manifest.json containing version, checksums, and a file list. The package includes everything in .pair/ — your customized guidelines, templates, and adoption defaults.
Distribute via GitHub Releases
Publish the package as a GitHub Release artifact:
- Tag a release in your central KB repository
- Attach the ZIP artifact from
pair-cli package - CI can automate this with SHA256 checksum generation
Teams install your custom KB by pointing pair-cli to your release:
Team Onboarding
When a team adopts pair with your organization's KB:
- Install —
pair-cli install --source <your-kb-url>downloads and installs the custom KB - Customize — the team adds project-specific adoption decisions (tech stack, architecture)
- Develop — every AI assistant reads the combined KB (org standards + team decisions)
- Update —
pair-cli updaterefreshes the KB when the organization publishes a new version
Version Coordination
CLI version maps to KB version. When you release a new KB version:
- Teams run
pair-cli updateto get the latest guidelines - Organization standards propagate automatically
- Team-specific adoption files are preserved (never overwritten)
Asset Registries
The CLI uses five asset registries with specific behaviors:
| Registry | Behavior | Description |
|---|---|---|
knowledge | mirror | Syncs KB content (overwrites stale files) |
adoption | add | Only adds new files (preserves team decisions) |
github | mirror | GitHub workflows and configuration |
agents | mirror | AI agent bridge files (AGENTS.md) |
skills | mirror | Agent skills distributed to AI tool directories |
The distinction between mirror (sync everything) and add (preserve existing) is key: organization guidelines update automatically, but team decisions are never overwritten.
Compliance and Governance
With a centralized KB, your organization gets:
- Consistent standards — every team follows the same guidelines
- Auditable decisions — adoption files and ADRs document every architectural choice
- Controlled updates — KB versioning ensures teams use approved standards
- Flexible customization — teams extend (never conflict with) organization defaults
Next Steps
- Enterprise Adoption tutorial (~60 min) — guided rollout walkthrough for organizations
- Understand how the Knowledge Base structure supports customization
- Learn how Skills automate the development lifecycle across teams
- Explore Agent Integration for multi-tool support